The Physiology of A Lepidopteran Muscle Receptor: III. The Stretch Reflex

The proprioceptive equipment of mammals is the most elaborate of any animal group. Nevertheless, muscle receptors not unlike the characteristically mammalian muscle spindles are also found among the arthropods, namely in decapod Crustacea (Alexandrowicz, 1951) and in three orders of insects (Finlayson & Lowenstein, 1958; Osborne & Finlayson, 1962). Lowenstein & Finlayson (1960) showed that the lepidopteran rquscle receptor (MRO) signals both phasic and tonic parameters of imposed movements, and in a quantitative study of the same sense organ (Weevers, 1966b) it was found that an individual MRO of a caterpillar provides information hardly less detailed and precise than does a mammalian spindle (Matthews, 1963). The present study was undertaken in order to discover whether the analogies between the two systems also extend to the manner in which muscular activity is controlled by stretch reflexes.

The arthropod reflexes which have received most attention to date are concerned with ‘startle’ responses. Thus Pumphrey & Rawdon Smith (1937) described the excitation of large fibres in the abdominal nerve cord of the cockroach by the action of air currents on the anal cerci, and Hughes (1953) described similar ‘giant’ fibres in the dragonfly nymph. It has since been confirmed that both of these systems are involved in evasion responses (Roeder, 1948; Fielden, 1960). Similarly, Wiersma (1947, 1949, 1952), among others, has described the physiology of the giant-fibre system of the crayfish. In each case activation of a sensory pathway, concerned with detection of environmental changes likely to culminate in noxious situations, stimulates a rapidly conducting interneurone system with a widespread motor field. Activation of these interneurones then results in movements appropriate to escape which override other current activities of the animal.

There is less information available about the functioning of those central nervous mechanisms of a less specialized kind which are involved in reflex adjustment of posture and locomotor activity. Pringle (1940) described depressor and levator reflexes in the cockroach leg. He also found that forced flexion of the coxo-trochanteral joint on one side produced slight but definite depressor inhibition on the other side. The sense organs involved were presumed to be campaniform sensilla, since pressure on appropriate regions of the cuticle evoked similar effects, whereas tension on various tendons did not. He was unable to show any effects on flexor-extensor balance in segments other than that stimulated. In contrast, Hughes (1957), who used amputation and transection of connectives in conjunction with cinematography on the same animal, had to postulate intersegmental connexions to explain his results. Wilson (1965) evoked this same stretch reflex using sinusoidal stimulation and found that on occasion the system could be driven to respond at 30-40 cyc./sec. Thus the duration of muscle contraction, rather than reflex properties, limits the phasic capabilities of the legs of cockroaches.

Although the studies of Bush (1962, 1963) were performed on decapod Crustacea and were concerned with reflex responses to stretch of chordotonal sense organs in the thoracic appendages, his results are very similar to those which will be described here. He showed the presence of negative feedback proprioceptive reflexes in a number of reptantian groups and analysed these more thoroughly than has yet been done for any invertebrate group. These reflexes are precisely analogous to the myotatic reflexes of mammals. The peripheral picture is of course considerably complicated in the Crustacea by the occurrence of neuromuscular inhibition. A large part of Bush’s studies was concerned with muscle antagonism and the respective roles of central and peripheral inhibition in the control of crustacean locomotion.

Other recently published reflex studies are less directly relevant to the present work and will only be briefly mentioned. Wilson & Gettrup (1963) described a proprio-ceptive mechanism controlling wingbeat frequency in grasshoppers; this is a slowly acting tonic reflex. Fielden (1960) and Mill (1963) demonstrated segmental reflexes in nymphs of Anax and Aeschna respectively following electrical stimulation of afferent nerves and Van der Kloot (1963) studied the reflex and other co-ordinating influences on the spiracular muscle of the pupa of Hyalophora cecropia.

Last instar larvae of Antheraea peryni were used almost exclusively in the present investigation. (In one experiment it was necessary to use the pupa). Experiments performed on larvae required the use of those which had not yet ceased feeding, since when the caterpillar starts to spin its cocoon it becomes more active, and consistent stretch reflexes are difficult to obtain. The feeding larva normally spends long periods motionless except for its jaws and head. The same was found to be true of acute preparations dissected in the manner previously described (Weevers, 1966a). Initially the sensory discharge from the MRO used to elicit stretch reflexes was monitored continuously. Later it became apparent that the afferent responses to a given stimulus were so consistent that there was no need for this. It was quite sufficient to check that the sense organ was functioning properly from time to time. The apparatus used to stretch the MRO in a controlled manner has already been described (Weevers, 1966b). The ‘stretcher’ mechanogram was displayed as a d.c. signal on the lower beam of the oscilloscope. The same beam could also be used to display muscle or other action potentials at the same time.

Muscle activity was recorded intracellularly with fine glass-insulated platinum wire electrodes of the type described by Ballintijn (1961). The rather obtuse-angled tip was extremely robust (one electrode lasting for many experiments) and did not usually penetrate through more than one layer of fibres, even when sufficient pressure was exerted to bend the thin shaft of the electrode. The Pyrex glass insulation was discontinued about 1 cm. from the tip, leaving a length of about 60 μ of the 12 μ diameter platinum-iridium wire core bare of the staff and springy coating. This gap acted as a pivoting point when the tip moved laterally. When the electrode was advanced so as to bow the long shaft, movements of the tip could be tolerated up to 4 mm. laterally and 1 mm. vertically without displacing the electrode from its intracellular location.

When it was desired to see the shape of an action potential relatively undistorted a coupling time-constant of one second was used in the preamplifier. Otherwise a 0·002 sec. time-constant was found more convenient, as it eliminated most movement artifacts cut down hum and showed the mechanogram signal to better advantage. Electronic apparatus was of the conventional type and has already been described (Weevers, 1966 a).

Unlike most other animals utilizing a hydrostatic skeleton, the caterpillar does not possess antagonistic circular and longitudinal muscle layers. Instead, bundles of longitudinally and diagonally oriented muscle fibres are interwoven in an extremely complex fashion; and in addition there is a discontinuous superficial layer of relatively short muscles (here called integumentary muscles) which are oriented in every direction possible. This adds up to a body musculature more complex than in any other insect. Lyonet (1762) described the morphology’ of Cossus ligmperda caterpillars. Part of one of his excellent figures is reproduced in Text-fig. 1. Because of its objectivity and brevity the same nomenclature will be used in the present work. However, to assist in visualizing the location of muscles being discussed, such descriptive terms as ‘dorsal longitudinal muscles’ may also be used where this is helpful.

Von Holst (1934) found that section of a single connective in the abdominal cord produced flaccid paralysis of the dorsolateral part of the segment immediately posterior and ipsilateral to the cord hemisection. Whereas stimulation of the integument in this region could still produce responses in other segments, no muscles could be induced to contract here by any form of mechanical stimulation. In the absence of more specific information, von Holst interpreted this as showing that the paired motor ‘centres’ for these muscles are situated in the ganglion of the next anterior segment. Kopec (1919) had observed similar effects but did not comment on them. Since pathways of this kind would appreciably increase afferent and efferent conduction times, electrical stimulation of motor and sensory axons was used, together with histological techniques, to examine this question.

As mentioned in a previous paper (Weevers, 1966c) the sensory axon from the MRO runs for a short distance separate from all other nerves. Extracellular electrodes could therefore be used to record action potentials in this single axon when it was stimulated electrically anywhere along its course. Owing to the polarized nature of all but a very few synapses, this means that one can map the path of the axon into the c.N.s. Text-fig. 2 a, shows the results of such an experiment. Measurements of the latencies of responses to stimulation gave, by difference, the conduction time between each of the arrows. Distances were measured with a micromanipulator having a vernier scale reading to tenths of a millimeter. The conduction velocity was nearly constant (between 0·8 and 0·9 m./sec.) along the greater part of the three branches of the axon. However, impulses travelled faster than 2 m./sec. along the anterior intracentral branch within the ganglion of entry into the C.N.s. No action potentials were recorded in the MRO sensory axon when stimuli were delivered to the contralateral connectives or to connectives anterior or posterior to the three ganglia shown in Text-fig. 2 a.

Text-fig. 2 b shows the results of a similar experiment, where a micro-electrode was used to record intracellularly from muscles while various points along the possible paths of the motor axons were stimulated. In this kind of experiment it is more difficult to achieve the same degree of certainty about the courses of the axons concerned than was possible with the MRO axon; since conduction is orthodromic, any synapse would transmit impulses and one might obtain responses to stimulation of interneurones. In fact stimulation further afield than the three arrows in Text-fig. 2 b did sometimes excite muscles, but the 1:1 relation between stimulus and response broke down at frequencies above 2o/sec. Furthermore, careful measurements of latencies revealed ganglionic delays over and above axon conduction time of 1-2 msec, in such cases. In contrast, it was not possible to show any ganglionic delay when recording from muscles responding in a 1:1 manner at over 100 stimuli/sec. ; such high-frequency following was seen only with stimulating electrodes located at the arrows in Text-fig. 2b. The conduction velocity along pathways of the latter type was around 2 m./sec., again being somewhat higher within the ganglion (2·3 m./sec.). This experiment was performed on many different muscles and it became apparent that only those innervated by nerve 2 behaved as though their motor axons passed along the ipsilateral connective from the next anterior ganglion. (Nerve 2 is the anterior of the two true segmental nerves, nerve i being a branch of the stomatogastric system.) It was also found, in agreement with von Holst (1934), that section of a single connective paralysed about two-thirds of the muscles in the half segment behind the cut. These were all innervated by nerve 2, and action potentials could not be elicited in such muscles by any form of sensory stimulation.

Further light was shed on the question of the course of the motor axons of nerve 2 by examination of stained sections of abdominal ganglia. Photomicrographs are reproduced in Pl. 1. The horizontal section in particular shows a tract of relatively large fibres running around the edge of the ganglion from the lateral part of the connective into nerve 2. These showed no sign of branching in any section. These are among the largest fibres in the cord, as can be seen in the transverse section. The parasagittal section shows the same bundle of large fibres once again, and in addition another bundle of (presumably sensory) nerve fibres of which only two are comparable in size to those in the first bundle. Since recordings showed the majority of efferent impulses to be larger than afferent ones, the tract of large fibres is almost certainly motor in function.

Text-fig. 3 shows two examples of the clearest and most typical reflex result of MRO stretch. The records in Text-fig. 3 a were taken from a tonically active muscle and those in Text-fig. 3 b from a muscle which was inactive at the time of stimulation. Both muscles were excited, most strongly during the period of stretching, and both showed some residual excitation after the MRO had been released. When a number of stretches were given successively, this residual excitation tended to build up.

The shape of the action potential is relatively undistorted in these records as a long time-constant was used in recording them. These two types represent extreme examples of the range of different shapes of action potential seen. Thus all muscles were of the ‘fast’ type with non-facilitating junctional potentials. Active membrane responses varied in size between the extremes seen in Text-fig. 3 a, b. Precise measurements of resting potential were not made, but from the change in potential on removing the electrode from its intracellular location, it was estimated that most action potentials overshot zero potential by about 3-6 mV.

Records taken from muscles in the segments adjacent to the stretched MRO revealed that many of these were excited also. In Text-fig. 4 such an intersegmental reflex is compared with the response of the homologous muscle within the stimulated segment. The two records were taken simultaneously. The intersegmental reflex was typically less intense but otherwise similar to the intrasegmental one. As far as this type of experiment could show, the latencies were little different. Certain contralateral muscle groups were also excited, group A for example, as seen in Text-fig. 5a.

Again the response was little different from the ipsilateral intrasegmental reflex, though perhaps the contralateral reflex was a little slower to reach peak excitation. One contralateral muscle, group P, was weakly inhibited by stretching the ipsilateral MRO, as seen in Text-fig. 5 b.

Inhibitory reflexes were decidedly unusual. Indeed, none of the longitudinal intersegmental muscles or long diagonal muscles were inhibited by MRO stretch. Unfortunately the integumentary muscles are rather inaccessible and only a few records were obtained. Some showed excitatory reflexes, but one in particular, group L, was strongly inhibited by stretch of either the ipsilateral or the contralateral intrasegmental MRO, as seen in Text-fig. 6.

It was shown previously (Weevers, 1966 b) that the MRO of the caterpillar provides the c.N.S. with very detailed information about the rate of extension, and even about the changes in rate of extension of a segment. It is of interest to examine the way in which this information is utilized. In fact phasic components can be seen in Textfigs. 3-5, but a low-frequency pulsatile discharge does not yield very good timeresolution; nor is Text-fig. 4 plotted in a way which would show rapid phasic effects. It was found by chance that embedding the appendages of a caterpillar in dental cement produced very intense and fairly constant muscular activity. In addition, motor activity could be increased by isolating a single functional ganglion from its neighbours. This procedure also rendered the discharge frequency more constant. The chosen muscle, group E, was also one which normally exhibited a tonic discharge. (Since group E is innervated by nerve 2, the smallest functional reflex unit was a pair of adjacent ganglia, owing to the path followed by axons leaving the c.N.s. in this nerve—see § (1).)

It may be seen from Text-fig. 7 that the reflex has a phasic component which, like the sensory response (Weevers 1966b,), is complex. The sense organ signals displacement, movement and acceleration and all three components appear in the reflex. The curves of Text-fig. 7 are remarkably close reflexions of the changes in the afferent discharge from the MRO.

Text-fig. 7 gives a good idea of the responses to stretching a single MRO. This sort of experiment is necessary for analysis, but the stimulus is a rather artificial one; in an intact caterpillar, displacements affecting only a single receptor must be unusual. Text-fig. 8 shows what happened when two receptors were stretched simultaneously. The ipsilateral and contralateral reflexes reinforced each other to give a response larger than the sum of the responses to stretching the receptors singly. Similar effects were seen when receptors on the same side in adjacent segments were stretched simultaneously, though in this case the facilitation was not quite as marked. No clear interactions were observed when receptors further afield were stretched at the same time as the ipsilateral intrasegmental MRO.

All the reflex data discussed so far were obtained using stretch-stimulation. This is ideal for evoking ‘normal’ responses but is not so well suited for investigating the synaptic mechanisms responsible for producing such responses. An attempt was therefore made to clarify the pattern sensitivity and complexity of the pathways involved in the stretch reflex, using electrical stimulation of the MRO axon. This technique is a difficult one to apply here because of the short length of single axon available for stimulation; in addition, the steady stretch discharge has to be blocked by pinching this same axon distally. These difficulties severely limited the data which could be obtained from available experimental material and also necessitated use of the pupa, where the length of single axon is greater.

Plate 2 illustrates results which were obtained using this technique. The record was taken from the dorsal nerve proximal to the spiracle and is therefore a complex one, but because of the low level of central excitability in the pupa, up to eight units can be clearly distinguished. In record 1, the smallest of the three units was the axon innervating the receptor muscle. This exhibited a tonic discharge which was briefly inhibited following the medium-sized spike passing centripetally along the MRO axon. The single large spike was in a motor unit showing an excitatory reflex. Other records yielded a latency for the earliest motor response of 30-33 msec. This seems a rather long latency, but it should be remembered that synaptic connexions are made in the next anterior ganglion. In one caterpillar, the conduction time over the same pathway was about 27 msec. Unfortunately it was not possible to identify in the time available which pupal muscle was excited when the above-mentioned reflex spike appeared, so it can only be said that the synaptic delay was probably of the order of 5 msec. The two stimuli seen in record 1 were the third and fourth in a train. There was no further reflex response after the fourth stimulus.

In record 2 the same units can be seen as in record 1. The large efferent spike usually did not appear until after the second stimulus, therefore the synaptic effects of the first MRO impulse persisted for at least 125 msec, and summed with the second impulse to produce a response. The pathway thus certainly exhibited properties of temporal summation and ‘fatigue’ or adaptation. It cannot be said whether temporal facilitation also occurred. Intracentral recording would be required to decide this. Record 3 again shows the effects of central adaptation, and two further efferent units can be identified. In record 4 where the frequency of stimulation was highest the state of central excitation was sufficient to evoke two spikes from an even larger motor unit, as well as some smaller ones, and the reflex was beginning to acquire considerable complexity.

It was mentioned in § (1) that only nerve 2 contained motor axons which originated in the next anterior ganglion. It was also shown that the MRO axon bifurcated in the ganglion where it entered the c.N.s. Intersegmental reflexes have been recorded in both the segment anterior and that posterior to the stretch MRO. The latter observation already implies reflex connexions within the ganglion where the MRO axon enters. It was of interest to examine whether the MRO axon also made reflex connexions with nerve 3 motoneurones in the ganglion where it entered the C.N.s.

Text-fig. 9 shows an excitatory reflex in a muscle innervated by nerve 3. When the c.N.s. was intact, excitation was of the typical phasic-tonic variety. On isolating the relevant ganglion from its neighbours the reflex, though reduced, did not disappear. In this case some of the characteristic features of the response appear to depend on intersegmental connexions (possibly non-specific), but there can be no doubt that synaptic connexions are present within a single ganglion. More cannot be said without finer surgical techniques.

Recordings of muscle activity were made from as many groups as possible in order to examine the extent of the reflex effects of stretching a single MRO. In the course of this investigation it was found that a few muscles, designated by Lyonet (1762) in his description of the muscular system of the goatmoth caterpillar as separate groups, did in fact share a common motor axon with another group nearby. The top three records of Text-fig. 10 show that for each spike in group H there was another in group G at the same instant, similarly for groups f and g and for θ₁ and θ₂. On the other hand, when recordings were taken simultaneously from pairs of fibres in ‘groups’ b and c, c and d, and b and d, only two motor axons could be demonstrated, not three. These two motor units will be called ‘bed prox.’ and ‘bed dist.’

Tables 1-4 show the results of about twenty experiments on last instar larvae. Recordings were made from as many muscle groups as possible, before, during and after stretching one MRO to 1·5 mm. above unstretched length at a rate of 0·2 cm. per second. This stimulus was repeated four times with a pause of at least 1 minute between stretches. The mean motor frequencies with the MRO stretched and unstretched were obtained from films of the results of such tests. An inhibitory response is indicated by a minus sign; all others were excitatory. No attempt was made to determine the statistical significance of reflex changes in frequency for several reasons :

1. The parameter of response measured tends to minimize phasic effects.

2. Several muscles were rather inaccessible and their motor axons were easily severed unknowingly during dissection.

3. In many preparations some muscles were difficult to excite at all, so that although when discharging they did show reflex changes in spike frequency, MRO stretch often did not initiate a spike train.

For reasons of experimental convenience all results discussed in previous sections and the data of Tables 1, 3 and 4 were obtained from abdominal segments, mostly the fifth. Accordingly Table 2 shows a selection of reflex responses to MRO stretch in thoracic segment 3 for comparison with the data of Table 1. The results show that similar stretch reflexes are present in the thorax, although they may be somewhat weaker. The reflex effects of stretch in any one preparation were usually very consistent, both in thorax and abdomen.

The tables contain a lot of data, so Text-fig. 11 was drawn in order to show pictorially the magnitude and location of each reflex. This figure is a diagrammatic representation of abdominal segments 4, 5 and 6 in a caterpillar dissected in the usual way. The most intense excitatory reflexes are shown by the most closely spaced lines, and inhibitory reflexes by dots. Records taken from muscles which are labelled but not shaded on this diagram showed no change in spike frequency when the MRO was stretched. Almost all of the other readily accessible muscle groups shown were also tested for stretch reflexes by listening for possible changes in spike frequency on a loudspeaker monitor. As may be seen from this diagram, when compared with the figure of Lyonet (1762) reproduced in Text-fig. 1, most of the major long muscles have been tested. Ipsilateral intrasegmental group B could not be examined with intracellular muscle recordings since this had to be removed to expose the MRO. I probably does respond reflexly to stretch, however, since extracellular hook electrodes on the nerve leading to this muscle showed the presence of at least three excited units.

Among the short integumentary muscles, only group P and group X are shown. If more muscles were drawn the diagram would become rather confusing; and in any case, owing to their inaccessibility only seven out of twenty such muscle groups have been tested as yet. Groups P and X are the only muscles in Text-fig. 11 which are innervated by nerve 3.

Present experiments provide strong evidence that motoneurones of nerve 2 in the abdominal ganglia of silkmoth caterpillars make synaptic contacts only in the ganglion anterior to the one at which their axons leave the c.N.s. This is in agreement with the observations of von Holst (1934) on Agrotis caterpillar. Hughes (1965) described a similar arrangement for certain cockroach motoneurones. In the caterpillar the MRO sensory axon divides into at least two branches, probably within the ganglion where it enters the c.N.s. The anterior of these branches passes up the ipsilateral connective to the next anterior ganglion where it makes synaptic contact with the motoneurones of nerve 2, mentioned above. Other sensory axons have a similar anteriorly directed intracentral branch. In the experiment of Text-fig. 2 a short latency responses following the stimulus in a one-to-one manner at high frequencies were recorded in branches of the dorsal nerve which never spontaneously carried centrifugally directed impulses. Such sensory neurons did not have a posteriorly directed intracentral branch as well as the anterior branch ; only the MRO axon bifurcated in this way.

Another anatomical observation of comparative interest emerged from examination of thin sections of caterpillar abdominal ganglia. A tract of large nerve fibres was clearly visible passing from the lateral part of each connective around the anterolateral margin of the ganglion and into the anterior part of nerve 2. In view of the path that they follow and in view of their large size these are probably the motor axons referred to above. If this inference is correct, then motor and sensory axons in the proximal part of nerve 2 are segregated into anterior and posterior bundles respectively. This is different from the dragonfly larva, where Fielden (1963) showed that the dorsal and ventral halves of the paraproct nerve roots were respectively motor and sensory in function.

Direct physiological evidence is needed to prove the existence of horizontally separated motor and sensory bundles beyond all doubt, but a further observation provides circumstantial evidence in favour of this idea. In the thorax of the caterpillar a nerve leaves the C.N.S. half way along the 1-2 and the 2-3 connectives. Appropriate stimulation and surgery revealed that this nerve innervated muscles which in the abdomen would have been innervated by nerve 2 of the ganglion immediately posterior. The nerve in the corresponding position to the abdominal nerve 2 appeared in the thorax to contain only sensory fibres. Thus the horizontal separation of motor and sensory bundles may here be complete. It is not known whether this arrangement has any functional importance, but it could be experimentally convenient (as the separation of motor and sensory roots has been in the vertebrates). Present observations likewise shed no light on the functional importance of the anterior displacement of the synaptic contacts of the motoneurones of nerve 2.

Text-fig. 12 shows the reflex pathways in the caterpillar which have been inferred from the observations discussed above. This figure also takes account of the spread of the stretch reflex as shown in Text-fig. 11. The number of synapses drawn is in every case the minimum which present observations would allow. Thus only the intersegmental reflex in muscles innervated by nerve 2 of the segment anterior to the stretched MRO has to be mediated via an interneurone.

The paired muscle receptors of the caterpillars are obvious candidates for a role in co-ordinating peristaltic locomotion. The pattern of muscular activity during crawling is complex, involving most of the body musculature in a co-ordinated sequence of contraction and relaxation. In this sequence major groups of muscles within a single segment act in an antagonistic manner (Barth, 1937)By contrast, the pattern of reflex changes following stretch of a single MRO (see Text-fig. 11), though widespread, is a simple one. It consists predominantly of excitation in muscles lying parallel to the stretched MRO—a classical myotatic reflex. Thus muscles which during peristalsis act successively or antagonistically are excited by this reflex to contract simultaneously. Therefore the timing of muscular activity during peristaltic locomotion cannot be rigidly determined by a reflex involving the MRO (cf. Weevers, 1965). These receptors function in a more basic role as the sensory elements in a negative feedback reflex adjusting muscular effort in relation to the load encountered. Intact crawling caterpillars also show such a resistance reflex (von Holst, 1934; Weevers, 1965).

It is of course possible that the experimental methods used here failed to reveal more complex proprioceptive responses of fundamental importance because no more than two MRO could be stretched simultaneously. Certainly, the spatial interactions between neighbouring receptors exert a profound influence on the intensity of the reflexes evoked. It can only be said that, in view of von Hoist’s (1934) finding that locomotor waves can pass as many as three denervated segments essentially unchanged, the pattern of muscular activation is probably largely centrally determined.

Thus the MRO of caterpillars fulfil a role in the reflex control of muscular activity analogous to the muscle spindles of mammals. It appeared from the work of Eckert (1961 a, b) on the fast abdominal flexor muscles of crayfish that the very similar muscle receptors in this group were used differently. But Kennedy, Evoy & Fields (1966) have shown that the slow abdominal musculature is controlled by a negative feedback reflex stabilizing bodily position and that the abdominal MRO are the sense organs involved. Although the receptor elements are different, similar reflexes also stabilize limb position in decapod Crustacea (Bush, 1963) and in insects (Pringle, 1940; Wilson, 1965). It seems probable that this control mechanism will be found to operate also in other groups of animals having similar sensory equipment.

I am very grateful to Prof. G. M. Hughes for guidance and encouragement during this work, and to the D.S.I.R. for financial support in the form of a Research Studentship.

Fig. 1, horizontal, fig. 2, transverse, fig. 3, parasagittal sections of a larval abdominal ganglion. The tract of motor axons which leaves the C.N.s. via nerve 3 is labelled M. The tract of (probably) sensory fibres in nerve 2 is labelled S (figs. 1 and 3). The lines in fig. 1 show the planes of the sections in figs, 2 and 3. The scale of the figures is shown on the left. Fixed in 0-2 % osmic in saturated picric ; stained in Heidenhein’s iron haematoxylin.

The stretch reflex recorded in nerve 2 of a pupa just proximal to the spiracle. The steady stretch discharge from the MRO was blocked by pinching the sensory axon where it left the cell body. The reflex was evoked by electrical stimulation of the MRO axon, distal to its junction with the dorsal nerve, at four different frequencies: (1) 1/sec., (2) 8/sec., (3) 23/sec., (4) 36/sec.